Contents lists available at ScienceDirect Catena journal homepage: www.elsevier.com/locate/catena A geostatistical sensor data fusion approach for delineating homogeneous management zones in Precision Agriculture A. Castrignanò a , G. Buttafuoco b, ⁎ , R. Quarto c , D. Parisi d , R.A. Viscarra Rossel e , F. Terribile f , G. Langella f , A. Venezia d a CREA Research Centre for Agriculture and Environment, Bari, Italy b National Research Council of Italy, Institute for Agricultural and Forest Systems in the Mediterranean, Rende (CS), Italy c Earth and Geoenvironmental Sciences department, University of Bari Aldo Moro, Italy d CREA Research Centre for Vegetable and Ornamental Crops, Pontecagnano (SA), Italy e CSIRO Land & Water, PO BOX 1666, Canberra ACT 2601, Australia f Department of Agriculture, University of Naples Federico II, Portici (NA), Italy ARTICLE INFO Keywords: Spatial data Proximal soil sensing Data fusion Change of support Factorial cokriging Precision Agriculture ABSTRACT Application of Precision Agriculture requires an accurate assessment of fine-resolution spatial variation. At present, advances in proximal sensing and spatial data analysis are available to characterize soil systems and detect changes in physical or chemical properties useful to understand and manage the variation within fields in a site-specific way. The objective of this work was to verify the suitability of geostatistical techniques to fuse data measured with different geophysical sensors for delineating homogeneous within-field zones for Precision Agriculture. A geophysical survey, using electromagnetic induction (EMI) and ground penetrating radar (GPR), was carried out at Montecorvino Rovella in the southern Apennines (Salerno, Italy). Both sensors (EMI and GPR) enabled the assessment of variation of soil dielectric properties both laterally and vertically. The study area is a 5 ha terraced olive grove under organic cropping. The sensor surveys were carried out along the terraces and over the entire field. The multi-sensor data were analyzed using geostatistical techniques to estimate synthetic scale-dependent regionalized factors. The results allowed the division of the study area into smaller areas, characterized by different properties that could impact agronomic management. In particular, a large area was delineated in the northern part of the grove, where apparent soil electrical conductivity and radar attenuation were greater. Through soil profiling it was shown that soils of the northern macro-area refer to deep, well developed, clayey Luvic Phaezem, whereas soils of the southern macro-area are shallower and less developed, sandy loam Leptic Calcisol. The proposed geostatistical approach effectively combined the complementary 2D EMI and 3D GPR measurements, to delineate areas characterized by different soil horizontal and vertical con- ditions. This within-olive grove partition might be advantageously used for site-specific tillage and fertilization. 1. Introduction Precision Agriculture (PA) is based on the assessment of within-field variation and, to facilitate the management of such variability, man- agement zones (MZs) are delineated. Management zones are homo- geneous sub-field regions with similar yield-limiting factors or similar attributes affecting yield (Doerge, 1999; Khosla and Shaver, 2001). Therefore, characterizing soil variation quantitatively and locally is crucial to accomplish the objectives of PA because optimum benefits on profitability and environment protection depend on how well land use and agricultural practices match local conditions (Buttafuoco et al., 2017; Castrignanò et al., 2000; Oliver, 2013). One of the greatest obstacles to implement Precision Agriculture derives from the difficulty to accurately determine local variation of agricultural inputs (Evans et al., 1996). An effective solution is offered by using real-time on-the-go proximal soil sensors to record soil data at fine spatial resolution (Adamchuk et al., 2004; Viscarra Rossel et al., 2011). There are already sensors of different type, which can measure soil moisture, micro- and macro-component contents, texture or other soil properties. Such sensors use a variety of measurement techniques (electromagnetic induction, electrical resistivity, ground penetrating radar and gamma sensors, multi- and hyperspectral spectroradiometer and fluorimeter) in conjunction with a global positioning system (GPS). Geophysical methods, in particular, such as electromagnetic https://doi.org/10.1016/j.catena.2018.05.011 Received 6 January 2018; Received in revised form 6 April 2018; Accepted 10 May 2018 ⁎ Corresponding author. E-mail address: gabriele.buttafuoco@cnr.it (G. Buttafuoco). Catena 167 (2018) 293–304 0341-8162/ © 2018 Elsevier B.V. All rights reserved. T